EBU R-128 meets K-System v2, a possible future for the loudness debate (Loudness War)

[Compyfox]

Wow... over 1500 reads since the opening ot this thread on 3rd January, but only 29 posts with barely anyone else further discussing this topic.

Did I literally kill my own thread yet again?

[paterpeter]

Did I literally kill my own thread yet again?

They are all gone remixing their songs @-16LUFS

BTW, it would be great to have some real world examples of the advantages of the additional headroom. For example, someone who does "professional" releases showing a mix/master of a track at "commercial" levels and a remix/remaster at K-16v2, making use of the extra peak headroom, of course.

[DuX]

Think of it as an educational thread, Compyfox. Everybody reads it, but they've got nothing to say. It is definitely an educational read. Good read everybody should read. There's really not much to discuss about it.

Since I mix at the -18dB RMS, I don't have to think about the headroom and peaks. That's one of its biggest advantages of mixing that way. Peaks never go beyond -6dBFS to -3dBFS at the most, so you can focus on the sound, not the master buss. I actually don't even look at it most of the time. And I love working with busses/stems, lots of them. They make mixing so much easier, and in Reaper it is so easy to setup separate busses.

[Compyfox]

They are all gone remixing their songs @-16LUFS

I sure hope so.

BTW, it would be great to have some real world examples of the advantages of the additional headroom. For example, someone who does "professional" releases showing a mix/master of a track at "commercial" levels and a remix/remaster at K-16v2, making use of the extra peak headroom, of course.

I unfortunately can not provide masterial myself. Also, reconstructing MP3s with "declipper" tools and posting it in here wouldn't be allowed either.

But... I can give you two popular examples of the Loudness War. One is Metallica's Death Magnetic again, A/B'ed with the official CD and the Guitar Hero (Metallica/GH4) release. And another (cross-linked) example with Billy Joel's "The Stranger" CD release (original, and remastered).


First Metallica's A/B:


Billy Joel's "The Stranger" CD release and remaster



Then two educational videos again:
(The Loudness War, Matt Mayfield, 2006)
(The Loudness War, Grant Leung, 2011)


And as prime example that a remaster can be pulled off well:
(A proper remaster, 2010)

Think of it as an educational thread, Compyfox. Everybody reads it, but they've got nothing to say. It is definitely an educational read. Good read everybody should read. There's really not much to discuss about it.

I don't know, I'm just... surprised. I have a feeling that KVR is not the right place for this discussion. Then again, 1500+ views tells me that there is just as much interest as with the Waves REDD thread, where I also added a lot of thread content.

Though more surprising is the fact that I didn't see this topic pop up on Gearslutz yet.

Since I mix at the -18dB RMS, I don't have to think about the headroom and peaks. That's one of its biggest advantages of mixing that way. Peaks never go beyond -6dBFS to -3dBFS at the most, so you can focus on the sound, not the master buss. I actually don't even look at it most of the time. And I love working with busses/stems, lots of them. They make mixing so much easier, and in Reaper it is so easy to setup separate busses.

The thing is, these so called "professional mixes" that are mixed in hybrid (ITB and OTB) are also howevering around -20dBFS/-18dBFS in terms of reference levels.

Though since there can be a lot done within ITB these days, and with one "make it loud" tutorial after another in every major music magazine, not to mention the industry still thinking "loud = more people will listen"... No wonder that this whole topic "loudness" went down the drain.

[DuX]

Speaking of mastering loud, I suppose you've seen this? It makes me laugh every time... What I absolutely cannot understand is that some people thought of it as being a serious tutorial. I mean... just look at some comments. Some of them are not sarcastic. People have completely lost their minds and ears. Sad.



We need an R128 "normaliser" kinda thing in every audio and media device around, so everything would sound as loud as everything else. We need that ASAP, no - yesterday, yesteryear... at the end of last century, actually. But better late than never, I suppose.

[BERFAB]

Wow... over 1500 reads since the opening ot this thread on 3rd January, but only 29 posts with barely anyone else further discussing this topic.

Did I literally kill my own thread yet again?

Perhaps I'm typical of many KVR'ers as far as skill level:

I know my way around a DAW and have at least a passing familiarity with the more technical aspects of recording. But I am NOT an engineer [and have only the utmost respect for those that are] which is why I buy just about every easy-tweak/beautiful UI/don't-worry-about-the-details plug that comes along. I read Sound On Sound and a half dozen other related publications fairly regularly, and I hang around here a LOT more than is probably healthy.

So when I see a well-written technical post like yours about one of the basic building blocks of modern recording, I like to think I can hang in there with it, despite my lack of formal technical training. But, if I'm being totally honest, I really can't. Oh, I get the gist: creating uniform standards may neutralize the 'war' and allow actual dynamics to become a part of modern recordings again. But to my eyes (paraphrasing of course), it reads like this:

"...blah, blah...compression...blah, blah...Metallica...blah, blah...over compression...etc."

About half way through, you note that a lot of bedroom producers will say that this stuff is "not important." While that may be true, I would humbly suggest that at least part of the reason is that the tools that are available [created by people WITH the proper tech backgrounds] are specifically created in order to REMOVE the technical burden from those end users. If and when more of those tools contain features that conformed to the proposed standards, those standards would, more often than not, be met.

Just a theory, of course. But my main point was to address why this isn't a longer thread. Your hard work is most certainly appreciated by many here. But it is at a technical level that is out of reach for many of us.

Much respect
-B

[No_Use]

Wow... over 1500 reads since the opening ot this thread on 3rd January, but only 29 posts with barely anyone else further discussing this topic.

Did I literally kill my own thread yet again?

For me at least, not at all.
I've read it with interested and fully sympathize with your efforts.
But, sorry to say, I'm a bit doubtful though if it will result in anything meaningful this time (I sure do hope so).
I mean we have the K-System, this German Tischmeyer thing and still I have to listen to awfully loud masters, bah.

But let's not be too pessimistic and hope for the best.

Btw, what let me wonder a bit, since you mention the maybe planned future implementation of auto-normalisation stuff (I think), do you know anything more about it ?
Is this going to be implement in playback devices like CD players, iPods etc. in the future ?

Will this be 'hard-coded' (ie. not possible to turn off) or did I misunderstand the whole thing ?

I mean I'd see where this is coming from, but then again it would also make me worry if this would be forced upon users.

[tanabarbier]

THANKS Compyfox!
Like a lot of people I am silently reading, taking the time to understand. Again thanks for the time and effort you put in this very educational thread.

Isn't that a crying shame that someone put a video on youtube to explain why sound quality is bad and making it a 240p or even 360p or 480p video? I mean, really?

[Compyfox]

Speaking of mastering loud, I suppose you've seen this?

First time I'm seeing this. Friggin hillarious. That fellow engineer has a nice SSL console, but dresses like a nerd and acts like a douche on purpose. Instant gold.

Reminded me a bit of the "mastering horror stories" videos on YouTube.

But better late than never, I suppose.

Like I said, I don't think it is done by introducing a loudness normalisation scheme on playback alone. The audio engineers and "bedroom producers" need to drive back on that behalf as well.

About half way through, you note that a lot of bedroom producers will say that this stuff is "not important." While that may be true, I would humbly suggest that at least part of the reason is that the tools that are available [created by people WITH the proper tech backgrounds] are specifically created in order to REMOVE the technical burden from those end users. If and when more of those tools contain features that conformed to the proposed standards, those standards would, more often than not, be met.

Your point is well taken. Yes, this whole topic is fairly technical. But much needed in my opinion.

We have so many tools at our disposal. But we don't really learn their purpose anymore. Just slam them on our material and be happy, maybe even call it a hit that went viral on YouTube. But then what?

Take a limiter for example. That device to me is still the last bastion I reach for - and only if I don't see any light of day. But most of the time, I don't need a limiter. The LAxA for example - I only use that one for artistic purposes these days (bass), not for signal limiting in form of a brickwall limiter.


We need more education again. And by more education I mean "the right directed" education. So much nonsense is floating around on the web, in magazines, transcended from hearsay, certain audio engineering schools, even certain personal opinions from well respected engineers around the globe (like "this has to be loud, that has to be overprocessed, etc). This is where we need to do a cut and start from scratch.

For the last couple of years that I'm on KVR, I try to be among the "educational" people. And I hope someday that it's paying off.

Just a theory, of course. But my main point was to address why this isn't a longer thread. Your hard work is most certainly appreciated by many here. But it is at a technical level that is out of reach for many of us.

I can agree on that end, but that doesn't mean that you can't educate yourself. Broaden your horizont a bit. Especially if you want to do everything ITB.

I learned it the hard way myself. I walked a lot of wrong paths, with crazy picked up ideas from tutorials, media, fellow engineers, etc. It took me a couple of years until I realised "wait a minute, something's wrong - you need to start from scratch".

If I had teachers back then (and not a "secret society") as I have at my disposal now, maybe something would have gone different. But then again, I wouldn't have learned a thing or found my own way.


I appreciate the feedback.

I've read it with interested and fully sympathize with your efforts.
But, sorry to say, I'm a bit doubtful though if it will result in anything meaningful this time (I sure do hope so).
I mean we have the K-System, this German Tischmeyer thing and still I have to listen to awfully loud masters, bah.

But let's not be too pessimistic and hope for the best.

Well, as stated both in the thread and in my white paper, both the K-System v1 and the DR-Meter by Brainworx/Tischmeyer have their flaws. To a certain extend, the K-System v2 as well. But with the EBU R-128 ballistics and weighting filter, we're at least closer to give a more... optimized loudness analysis.

At least IMO.

Btw, what let me wonder a bit, since you mention the maybe planned future implementation of auto-normalisation stuff (I think), do you know anything more about it ?

Only what is officially stated by the "Music Loudness Alliance" (see page 1), and what I heard from the people that are involved (which was pretty much the same).

Is this going to be implement in playback devices like CD players, iPods etc. in the future ?

That is the plan, yes.

Will this be 'hard-coded' (ie. not possible to turn off) or did I misunderstand the whole thing ?

Also to my understanding, it will be hard-coded. You can still change the volume, but what happens behind the scenes is hard to explain (please consult the white papers of the "Alliance").

The thing has savety mechanisms however. Or are at least planned to have some. For example, a certain volume can not be exceeded, and you need to confirm it several times to go there (on iPod's for example). This is to protect your ears from damaging. And I think this already starts at 75dB and ends way before 90dB. This is IMO a good idea.

Again, what's happening on playback is an own topic in itself, so please consult the white papers of the "Music Loudness Alliance".

I mean I'd see where this is coming from, but then again it would also make me worry if this would be forced upon users.

I also share the worries, but nothing is set in stone yet (first and foremost). And to my understanding, it should be less intrusive than forcing users to go through iTunes or staying constantly online in order to access certain services of your favoured tools.

Think of it as dynamic presets in a DVD player that you can setup for playing back a movie with DTS audio stream that is a tad too dynamic for your liking. Granted, a totally different topic, but it's also not intrusive. At least not IMO.

I actually welcome such a scheme (loudness normalisation). I can then concentrate on turning the volume up and down to my mood, rather than constantly reaching for my smartphone to adjust the volume for each track that is playing. And that at -20degrees on the way to the subway station. (nope, no remote for custom headphones)

THANKS Compyfox!
Like a lot of people I am silently reading, taking the time to understand. Again thanks for the time and effort you put in this very educational thread.

A pleasure.
And it's good to see that more people chime in and respond rather than stay "read only".

Isn't that a crying shame that someone put a video on youtube to explain why sound quality is bad and making it a 240p or even 360p or 480p video? I mean, really?

I think you mean the Billy Joel comparision. Yes, not the best quality, but I think the point came across.



Again, thanks for all the feedback and discussing the whole topic. Definitely appreciated. This spreads the word.

[No_Use]

Thanks Compy for taking the time to answer in depth.

[Sounddigger]

Wow... over 1500 reads since the opening ot this thread on 3rd January, but only 29 posts with barely anyone else further discussing this topic.

Did I literally kill my own thread yet again?

Think of it as an educational thread, Compyfox. Everybody reads it, but they've got nothing to say.

People probably have a lot to say, but are understandably hesitant (as I have been) since this is a pretty technical subject dealing with some new ideas, and you really have to get some hands-on experience to make a good evaluation of the various propositions.

I think this is a great thread! Levels and dynamics are such an important topic!! If melody, timbre, and rhythm are three dimensions of music, than loudness and dynamics are definitely the 4th. It's great to keep thinking about this subject. I've read through quite a few threads on the subject in this forum, and would especially like to thank Compyfox for some very nice work that can be found in them.

While I was aware of Bob Katz's anti-loudness activism, I was not yet familiar with his K-metering system, and Compyfox's explanations of how to set up the PSP vintage meter has proven to be really very useful to me, both for recording and mastering at the right levels (see topic found at http://www.kvraudio.com/forum/viewtopic.php?t=178557 ).

I have struggled with the loudness aspect of recording and mastering for quite a long time. When I started recording my music, the loudness war had yet to begin. Just adding a bit of compression here and there was enough for delivering a good mix for broadcasting, which was my main field. Then suddenly, it became common practice to squash the hell out of audio (especially for TV and radio commercials) and when I went on mastering my own work I made some big mistakes, especially thanks to excessive use of Waves' L1 maximizer. Good expert advice in this field was not yet easy to come by... I took a brake from recording for over a decade, and over the years I've been finding all the right answers to these issues, especially thanks to the internet. Wonderful!

One thing I'd like to point out is that the term "loudness war" is quite a misnomer, the way I see it. The reason is that in reality, extreme compression and brickwall limiting are not really making music better when it is heard loud, and it's not really making it louder in a strict way either, i.e. it's not taking the audio beyond the full range of available dBs, as people here will be aware of.

As far as I can tell and my experience goes, the main usefulness of extreme compression and limiting is to make music sound "loud" when it's auditioned at LOW levels, or to make it feel louder than it really is when you can't raise the volume due to some restriction, such as legal restrictions (as can be the case for night-clubs), or as is the case for broadcasting. And it has to be appreciated that as good as a truly dynamic mix sounds when it's played at a loud level (say around 83 dB SPL and above), when played at low to very low levels, it WILL actually appear to sound "louder", punchier, and (arguably) "better", with appropriate limiting/compression (and with still a fair DR). After all, the main purpose of limiting and compression is to even out the dynamics, which is advantageous for overall presence and for experiencing various musical information the music contains at (excessively) LOW levels.

To make music that has been limited to smithereens sound good when played loud, it has to be played TOO loud, by which I mean: louder than would have been necessary if it would not have been limited to death. It's only when being played TOO loud (with lows and body of sound louder than non-compressed version), that some of the sensations (i.e. such as the artifacts) of the original dynamics are (somewhat) experienced. Still, even when played too loud, such music can be conforming to legal or structural limitations where non-compressed/-limited music wouldn't, which can be an advantage.

Some people interprete the effects of over-compression/limiting in a way that is all mixed up, and are led to believe that music that has been excessively limited/compressed sounds "dull" once the volume is turned down. But that depends at which level of SPL you call a volume "turned down". My experience is that at VERY low levels (equivalent to someone speaking at very low volume, almost whispering), excessively (but expertly) limited/compressed music sounds "better", because it has more "presence" (generally speaking). This makes sense, really. Just try it out. At some intermediate volume levels the same can be true, depending on various circumstances.

For this reason, there is something to say in defense of mastering with extreme compression and limiting, in particular when tracks are meant for being played at (very) LOW levels. Typically, a release for radio would be such a legitimate reason, although it can be argued that it's better to leave compressing and limiting to the station instead of making it an integral part of the released version; and now we have the EBU R128 which aims at making the practice entirely obsolete for (most) broadcasting.

While it's presently more and more common to condemn the excessive (and deteriorating) compressing and limiting that we see/hear today, you have to consider that most of the world listens to music when it's played at low levels, most of the time. Think about all those listening to music on their PC speakers, those in their cubicles, those in supermarkets or shopping malls, those listening to a radio in the background, in the car, a (mediocre sounding) TV in the living room or bar, an iPod, etc..., you get the idea...

This is one reason why extreme compression and brickwall limiting can work to no uncertain extent. I.e., it can work to catch the attention and to sell tracks to an audience that is listening to the music at LOW levels!!! So really, this is about how the "LOW level" war can be won!

Some people have looked into the question if excessive limiting makes certain tracks more successful than others, and they found this not to be the case. But I beg to differ, since these surveys typically concern tracks that have already gained attention to a certain extent. Comparing the commercial success of tracks made by artists who are already successful is one thing, comparing tracks from artists who are complete unknowns is quite another. However, I'm not claiming that excessive limiting is the key to a HUGE success, it can only play a key role in getting tracks noticed I believe, that's all.

Sometimes I get the idea that the discussion about the "loudness war" is mostly held by people who only hear music on pristine monitoring systems, in conditions that are top-notch for audio, and without any neighbors that could complain about loud music... The truth is that more often than not, music-recordings get abused in all sort of ways, and only part of the world's musical experience consists of hearing music-recordings on (fairly) good audio systems, it be at home, in a club, or in a cinema.

The new EBU R128 standardization is a big step. And the standard definitely makes much sense, in particular for broadcasting programming - although I believe it should be left to every station to use it as it sees fit, and not be obligatory.

I haven't seen it mentioned here, but for people who have missed it, there's a very good example that clarifies the whole concept , as given during a presentation by Florian Camerer (mentioned in the first post). It nicely shows the difference between traditional peak calibration/normalization and the average calibration obtained with the new standard:

Video: EBU R128 Introduction - Florian Camerer (full)

see/hear the comparison at 37:50

Or see it in part 3 of 4

with comparison at 14:18

While the new EBU R128 makes a lot of sense for broadcasting programming, I believe it's much less relevant for music production, and even quite impractical.

For me, since I'm a composer/musician/producer, my main objective with this whole subject is to decide on how I will be leveling my tracks. I don't really have much interest for the technical side, apart from the practical application of things.

I believe there's a place for every kind of approach, including for "over"-compressing/limiting tracks. Actually, I love the sound of a pumping compressor, when used appropriately, it be on individual tracks or on a complete mix! If someone wishes to produce an uber-compressed master, then I'm against some kind of audio-police turning down the overall volume on the medium (CD, DVD, download) it is released on. Before anything else it's an artistical or conceptual choice, and it should be free for anyone to decide on.

Many recordings of classical music, especially of symphonic works, are extremely dynamic. Frankly, I often find them TOO dynamic for my taste, and will be found regularly adjusting the volume when listening to such recordings. Such a dynamic range may be great from an audiophile perspective and from a concern to render the original sound source as faithfully as possible. But I believe this is not reckoning enough with the differences between listening to a symphonic work in a concert hall and through a pair of speakers in circumstances that can vary greatly.

My experience is that even many movies and other film or TV productions can benefit from more compression/limiting, not less. The reason for this is what I already described: there are times you're trying to view a movie/TV/video production at a LOW volume level, when it can be really annoying if the dynamic range is beyond a certain span. Those who have tried to view a movie/TV/video production at night in a room where the noise could disturb and/or wake-up others in the house, might understand what I mean. That is NOT the time to go EBU R128 -23LUFS!!

As a matter of fact, I regularly listen to all kinds of internet radio talk-shows, and for the reason just described, or during certain activities where there is background noise to deal with, I almost always squash such audio into an almost perfectly flat loudness curve (well, sorta), which is a piece of cake in Winamp using the nifty "Loudmax" maximizer plugin (which is also available as a VST, by the way - excellent for squashing voices, not so great for mixes rich in bass, find it at http://loudmax.blogspot.fr/ ). Occasionally, I might do the same with a classical music radio station, albeit to much a lesser degree.

In the video "EBU R128 in Transmission and Production", a presentation by Thomas Lund, an interesting graph can be seen showing different loudness range targets for different circumstances, which illustrates my point to a certain extent, as related to background noise and circumstances. See at 39:40

I think there simply isn't one and one solution only, that fits all. Not for broadcasting, and much less for music recording.

On the other side of things, for my own productions I definitely aim to record and mix with the best dynamic range possible and appropriate. So I have been testing all the ideas I found, including the suggestions made in this forum.

It seems to me that the Bob Katz system, especially the "K-14" (v1) setting, is the most useful for achieving a dynamic mix, across all genres. I don't see why it should be considered "obsolete", since it works so well, even if not tailored to the new EBU R128 standard.

In order to test things, I went out to look for good "benchmark" examples, in a way similar to what Compyfox has been reporting in the previous post.

The most dynamic and interesting mixes of "high-energy" music I could find in my collection were those made by Bruce Swedien for Michael Jackson.

I recommend anyone interested in metering to check them out while looking at a few good meters. Two most amazing mixes are definitely found on MJ's album "Dangerous", which I took from a CD published in 1991. Also see the following video versions (but preferably check a CD/wav version on the meters in your DAW of choice):

Michael Jackson - Dangerous - Jam



Michael Jackson - Remember the Time


When you look at a track like "Jam" with the TT Dynamic Range Meter, you'll find that the DR range is at an averages of around 15!! In the loudest parts, it hardly gets under 12!!!
(for more stats, also see http://www.dr.loudness-war.info/details.php?id=31923 )

As a comparison, the winner of the "Dynamic Range Day Award" 2011/2012 (see http://dynamicrangeday.co.uk/award/ ) was Björk's "Biophilia", which has an average DR span of 8, and a maximum DR span of 11 (as checked on http://www.dr.loudness-war.info/ ) .

The TT Dynamic Range Meter might not be the ideal tool to measure a track dynamic "non-loudness" quality, but I like it better than the "crest factor" metering as found on Voxengo's Span, and it does give some good indications.

It's not surprising that Bruce Swedien's mixes turn out to be so dynamic, as he is known to use NO compression and limiting at all, or only exceptionally "just a squirt", as he puts it himself.

Compared to other tracks, if you lay these mixes side to side A/B-ing them with BS's mixes, then most of them will sound dull and lifeless to no uncertain extent. The more they are compressed and limited, the more they will sound "dull" or muffled (when heard at a normal non-earshredding volume).

As BS puts it in a video interview:

"Compression is for kids"

LOL!

See the following video, at 16:06


Here are some other interesting quotes from Bruce Swedien:

From http://www.soundonsound.com/sos/nov09/a ... wedien.htm

Given Swedien's repeated emphasis on maximising pickup of transients through mic selection and careful use of the recording medium, it makes sense that he has trenchant views on the use of compression. "I'm not a big fan of compression or limiting at all — I can't emphasise that enough. On many of the recordings that you hear today, all the excitement and all the colour is gone because they're so over-compressed. I never did that. I would never have a compressor or limiter on the [master] bus, for instance. I want all that transient information there. And no compression or limiting on any drums or percussion. That's one of the biggest mistakes that I hear, I think, in modern pop recording. The stuff is so compressed they've limited the living doo-doo out of the sound."

That's not to say that he leaves the dynamics of the performance completely untouched, but is much more inclined to achieve the required dynamic control through to-tape fader rides while overdubbing and automation while mixing. "I'm a nutcase about details in the mix, so I'll use automation to a degree, but only very subtle compression. I have a pair of the new variety of [Universal Audio] LA2As that I just love, so I will use those, but it'll only just be tickling the meter, at the most one or two decibels. I don't like what happens to the sound when you compress any further, and that's very important to me."

If you're checking BS's mixes of "Jam" and "Remember the Time" with the EBU R128 standard, then it appears these masters would have to be turned down respectively by 9.8 and 13 dB, if my understanding of the standardized leveling method is correct!! That's obviously quite ridiculous, especially considering how terrifically dynamic these mixes are (i.e. with excellent DR/crest factor).

I haven't yet checked this with TC Electronic's LM2 meter (which seems the most authorative), but only with Tonebooster's EBULoudness meter in the "LFS EBU R128" setting, by aligning the peaks on the SLk's -23 LUFS level. The way I understand Florian Camerer's explanation of the standard, it's not necessary to align the peaks at MLk -23 LUFS, the idea is to find an average level that works over time. Peaks are permitted, and even welcomed. But maybe I'm applying the -23 LUFS rule incorrectly here. Anyway when aligning peaks on the MLk's scale this would result in even greater "corrections"!

Actually, due to the lack of clear instructions, it proved to be quite difficult to calibrate music tracks to what is supposed to be the EBU R128 norm. There doesn't seem to be an absolute method for it (as is pointed out repeatedly by Florian Camerer in the video named above), but aligning peaks at the SLk's 0 level in the "LFS EBU R128" setting appears to get the right results.

... For compensation, I mostly trusted my instincts and went for values that were around "ML average" values.

Unfortunately, you can not measure "avg" values for ML as of this moment. So I hope to see an update for TB Loudness where I could see a histogram for the ML as well instead of SL alone. This makes it easier to judge where the average level of the production was.

But nothing that can't be done through training and trusting your ears.

Compyfox, I don't really understand your method of deciding on how to "correct" the levels of the releases you checked. It seems still pretty arbitrary, notwithstanding the very technical approach. If you want to get as rigorous as you aim to be, why not stick to aligning peaks on the SLk's 0 level (with LU K14v2 setting) or to LM2 read-outs (or something based on that meter)?

Also, it seems you're not taking into account that digital releases of music tracks are generally normalized (at least those from the 1990s on). If done properly the normalization would be with True Peak at 0 dBFS, but - as you found for certain tracks - more and more people seem to want to cheat by normalizing beyond it, with TP at up to +3 dBFS!! Ridiculous, and pretty silly, really.

Much of the corrections you advocate would be much reduced simply by taking this normalization into account (and removing it in the "correction"). But normalization is just an issue of optimizing the digital medium, and not allowing it would mainly result in downgrading its use. Is that worth it?

Again, my only concern is what guidelines to follow to realize a mix with a nice dynamic range, and what lessons to take from all this. When looking at other mixes/masters, it seems to me that if you want to get an idea of what the mix really looked like on the meters during mix-down, you have to take off a few dBs from the released version to come back at the original level.

When looking at Bruce Swedien's benchmark mixes, you have to consider the fact that no processing at all has been done after HIS final mixdown, excepted probably a slight normalization to put the TP at 0 dBFS. On my meters the TP for the tracks I mentioned is generally actually slightly above it, at +1 dBFS.

Bruce Swedien emphatically states that after his mixdown, nothing is done to the track at all, meaning NO mastering engineer gets to touch his tracks once BS is done! (see a forum topic he participated in at http://www.gearslutz.com/board/bruce-sw ... s-mix.html ). He is surely as truthful as he can be about it, yet this must be taking things without the normalization. I mean, he claims not to use his meter(s), at least not for making final decisions (see link), but I simply can't believe he would manage to mixdown with True Peak exactly at 0 to +1 dBFS!

Also, if you look at a meter specified to the K-system, BS's mixes at first are hard to put into either the K-14 or K-12 category. But this changes completely if you lower the level so that TP is at around -3 dBFS. It seems very logical that that he would mixdown leaving 3 to 2,5 dB headroom. Once you lower the level with -3 dB, you get a TP at around -2 dBFS and you'll find that his mixes read out beautifully on both a K-14 and K14v2 meter.

The same trick seems to give good results to analyze other tracks. After correction and setting TP at (around) -2.5 to -2 dBFS, it becomes really clear if a mix must be seen as fitting K-14 or K-12.

(Incidentally, since BS's mixes have no compression and limiting at all (or hardly), it's ideal material to experiment with how compression and limiting work out at different volume levels/SPL on an outstanding mix! ).

It seems to me that the "LU K16v2" setting is too strict for "high energy" mixes, and even for most mixes of classical music. Calibrating Bruce Swedien's mixes of "Jam" or "Remember the Time" with the "LU K16v2" setting with peaking on the SLk's 0 level, would mean they would have to be turned down respectively 5 and 6 dB, and when aligning peaks on the MLk's 0 level, it would mean they would have to be turned down respectively 8 and 7.5 dB. A tad excessive maybe?

Even the recordings of various symphonic works I checked would have to be leveled down substantially, while their dynamic range is extreme when comparing soft parts with louder ones. Surely, this can't be what is desired, seeing how dynamic these recordings are. And how would you master two consecutive tracks on a CD, where the first track would be a full blown symphonic work with FFF passages, and the second track would be a piece for harpsichord? (This is an example I found on one of my CDs) Would you insist that the harpsichord track is leveled at a fraction of the volume of the first track? This is what a faithful reproduction of dynamic range would dictate... However, any reasonable balance between the two tracks will be a fictional/virtual-reality one...

So to conclude, I don't think it will be useful at all to impose a "LU K16v2" standardization for musical releases, much less even practical . And Compyfox, you even mention in your first post in this thread that you wouldn't use the K-16v2 setting to mix popular music like rock/pop/electronic. So what's the point of demanding that it become a universal standard for musical releases?

Another example is Bob Katz, who recently switched from using his own K-System to using configured meters of the EBU R-128 standard. But as with the first introduction of the K-System, his ideas are still pushed aside as "utter nonsense".

I don't understand at all why Bob Katz would completely switch meters. I really like the 600ms integration time setting of the original K-system. A K-14 meter with 600ms integration time suits me perfectly for just about any master bussing (together with DR metering and keeping an eye on True Peak). I find the 400ms of the K-14v2 MLk a bit too frantic to stay in Zen mode.

Could you expand on what some significant advantages could be of a 400ms setting, for tracking/mixing music, aside from the fact that the EBU R128 standard is based on 400ms?

Do I understand correctly that in the "LU K16v2" of Tonebooster's EBULoudness meter, a Hi-Pass filter is used? Seems interesting, but is it really the best way to represent the dynamic range of a audio signal? It might be better adapted to the human ear, but does it take enough into account the enormous differences that exist between the systems that recordings are played on (i.e. ranging from top-notch calibrated systems to boomy and/or muffled to thin and shrill).

Btw, what let me wonder a bit, since you mention the maybe planned future implementation of auto-normalisation stuff (I think), do you know anything more about it ?

Only what is officially stated by the "Music Loudness Alliance" (see page 1), and what I heard from the people that are involved (which was pretty much the same).

It seems to me this is the only practical solution to the whole loudness issue. But I would only opt to buy a device where there's a button to turn the normalization off!!

Since I mix at the -18dB RMS, I don't have to think about the headroom and peaks. That's one of its biggest advantages of mixing that way. Peaks never go beyond -6dBFS to -3dBFS at the most, so you can focus on the sound, not the master buss.

There are different ways to measure RMS too, and there are differences between meters in how they represent RMS, using different speed settings. Some meters give a "continual" RMS reading, with fast integration time. DFX RMS Buddy has this, and it appears that the TT Dynamic Range Meter readings for RMS correspond to this more or less (makes sense really, as it's used together with absolute peaks for measuring the DR).

You can also use the "integrated loudness" readings on the Tonebooster EBULoudness meter, as this is equivalent to RMS (though not exactly the same), as described in the article at http://www.soundonsound.com/sos/sep11/a ... udness.htm (note that the IL readings on the TB-EBUL meter don't respond to volume changes though, different to normal RMS).

Anyway, lots of technical stuff here to get distracted over, so to quote Bruce Swedien: always think about the music FIRST!

Sorry for the long post. Hope it is of interest to some

[Sounddigger]

While browsing and digging further into this subject, I got to the "EBU Tech 3343" document, in which is described how the loudness normalization/calibration is to occur.
It can be found at
http://www.tcelectronic.com/loudness/br ... standards/
or at the following direct link:
http://tech.ebu.ch/docs/tech/tech3343.pdf

It also contains many of the graphs and text bits seen in the videos mentioned in this thread.

The graph illustrating the background noise problem, can be found on p. 19.

On p. 23 and onwards, a description is given of "Loudness Metering for Production and Post-Production" (chap 4.2).

It reads:

(p. 23) The M and S time windows are intended to be used for the immediate levelling and mixing of audio signals. Initial level setting may be performed best with the Momentary Loudness Meter, adjusting the level of key or anchor elements (such as voice, music or sound effects) to be around the Target Level of -23 LUFS. Of course a mixer has to know at any time how loud the actual signal is, and that is the main purpose of the Momentary and Short-term measurement [emphasis mine]

So if I interprete this correctly, the Tonebooster's EBULoudness' MLk / 400ms meter should only be used to set an initial level, not as an overall peak level. The SLk / 3s meter logically gives more of an average loudness level, and it seems to me that for many kinds of music this is a fine time span to set a level that will result in the desired "anti loudness-war" effect in the spirit of the EBU R128 concept.

[DuX]

I've been with you until you said "I believe it should be left to every station to use it as it sees fit, and not be obligatory", and further on.

Thank you for such a profound article, but what you said wouldn't work. It should be enforced on every station. It should be enforced in every media player there is, too. Because people are dumb and don't know, nor perceive, what we know. I know that a louder master doesn't make it great, or actually more like sounding like piece of cra*. I know that loud doesn;t make a piece of music better than someone else's piece of music that sounds less loud.

However, so, common people might like a song from somebody who recorded it, and mastered it at K-14 levels. If you put such a recording against an "overloud" recording at the *same level* they would probably not like what they hear in the louder recording. People's perception have been skewed for years now and they don't know what a proper recording sounds like. This would level the field for everybody. That's why I'm so FOR IT. If your music sounds good, it will succeed. No matter the loudness level! It is a well known fact that louder sounds better to common people, and by levelling the loudness level of reproduction, we would take out the miserable "loudness factor" that nobody understands [but hears] out of the whole equation. Then, if you like the distorted to hell and "loud" banana songs I've been hearing lately, it's yours to have, at a lower level, and on the other hand, if you like what some "Joe from Texas" has done at his home, recording at K-14 levels, you can have that, too, at the *same* level.

I call that FAIR. To say the least. And it will start a new recording revolution - quality over loudness.

Cheers!

[DuX]

When I listen to music, I use Foobar2000 with replay gain on, all the songs being R128 levelled, and I can tell you I find it really hard to believe anyone would prefer the music that comes at the loudness levels these days. It sounds flat and just pitiful. Like listening to an AM station in comparison to older recordings from 80s and earlier.

[Compyfox]

A lot of wood to chop through, and I definitely appreciate the posts so far. But I need a bit to answer properly. Still recovering from a virus.



Just a couple of things beforehand:

a) The K-System v1 is an unweighted RMS meter, setup to 600ms with a shifted reference level. The K-System v2 is built upon the ML ballistics of the EBU R-128/ITU-R BS.1770-2 standard, weighting filter (lowcut, highshelf) prior to measurement, signals summed, ballistics 400ms, also a shifted reference level. The available DR-Meters (which are crest factor meters btw) use no weighting, though bx_meter offered additional weighting filters.



b) I chose the ML meters after heated debates with Mister Katz (and also after his recommendation), since it's the most closest to the old K-System ballistics. I see a reason in using the SL meter for actually "checking" loudness over time. But depending on the production, and with the 3s measurement timeframe, chances are that you get very different readouts. This is why I say "focus on ML". So unless all the available EBU R-128 meters offer histograms for ML, you can indeed focus on SL (which records the histogram) for the time being, but don't go higher than +1,5LU. That is also mentioned in my white paper.

Though focusing solely on the SL meter and applying the same rules for it than with ML, could mean a shift of the ML meter in terms of loudness spikes. So... trust your instincts, and what the meters give you as readout. Nothing much changed, only the ballistics did. Keep in mind, we "abuse" the EBU R-128 meter for measuring music - and for this, the ML meter is definitely suitable.



c) I do not use an EBU R-128/K-System v2 meter while recording/mixing. For this I use a VU (300ms, -18dBFS, unweighted)/Digital Peak meter combo. For mastering I use the EBU R-128/K-System v2 meter to judge a final loudness.



d) TP of -1dBFS maximum has a reason in this particular case (for music) as well. It's a savety mechanism. MP3 and AAC encoders have a certain limit (especially a high loudness) that is even worse than old DAC's. So in order to retain a good audio signal, stay below or at -1dBFS maximum. As reference, I can recommend you to take a closer look at the SONNOX Fraunhofer Pro-Codec video presentation on YouTube (see )



e) I never said "I wouldn't use K-16v2". At least not to my knowledge currently. I do however suggest a certain timeline in my white paper (page 9) to gradually go down from K-5 to K-16v2. I am not enforcing it, I am suggesting it to get back to the quality of the early CD days.



f) I can somewhat agree with the "noisy environment" thing and that a proper limited stream does work better in this case(!). But in order to do so, the overall tracks need to be aligned first, then maybe further processed. Else we still have jumpy/pumping audio streams. This is however not our (the musician's and mastering engineer's) concern as of this moment. This is a concern for the broadcasting stations. Or later in playback devices, the Loudness Normalisation schemes.



g) Our main focus as musician and engineers is to mix as dynamic as possible. In case of audio engineers, the lower the loudness of the endproduct (CD, MP3, HD Audio), the more enjoyable it is, the easier the load of broadcast compressors will be, the less noticable volume jumps/pumping you will get. Not counting the benefit of switching disks on a DVD player from a movie to a CD soundtrack, without touching the volume fader.

Broadcasting stations over here in Germany generally run a normalisation process to turn down(!) the volume prior to sending the content, rather than turning the volume up. Again, this is to ease the load of the compressors.




Our focus (audio engineers) should be to provide the most clean, most dynamic and non-squashed to bits production we can pull off. Everything else can be handled with presets (Dynamic Rage presets in DVD Players), possible future Loudness Normalisation schemes in playback devices or by the broadcast environment.

Apparently, a lot of well known engineers in the industry already do that (or do it again). But why it's turning into crap while mastering is still a mystery to solve.




I hope that counts as an answer for the time being.